We present a measurement of the polarization observed for bottomonium states produced in p-Cu collisions at square root of s = 38.8 GeV. The angular distribution of the decay dimuons of the Upsilon(1S) state shows no polarization at small values of the fractional longitudinal momentum x(F) and transverse momentum p(T) but significant positive transverse production polarization for either p(T)>1.8 GeV/c or for x(F)>0.35. The Upsilon(2S+3S) (unresolved) states show a large transverse production polarization at all values of x(F) and p(T) measured. These observations challenge NRQCD calculations of the polarization expected in the hadronic production of bottomonium states.

Optical-window birefringence is frequently a major obstacle in experiments measuring changes in the polarization state of light traversing a sample under investigation. It can contribute a signal indistinguishable from that due to the sample and complicate the analysis. Here, we explore a method to measure and compensate for the birefringence of an optical window using the reflection from the last optical surface before the sample. We demonstrate that this arrangement can cancel out false signals due to the optical-window birefringence-induced ellipticity drift to about 1%, for the values of total ellipticity less than 0.25 rad.